|Publication number||US7661107 B1|
|Application number||US 09/484,549|
|Publication date||Feb 9, 2010|
|Filing date||Jan 18, 2000|
|Priority date||Jan 18, 2000|
|Also published as||US7987465, US8381223, US20100122262, US20110283293|
|Publication number||09484549, 484549, US 7661107 B1, US 7661107B1, US-B1-7661107, US7661107 B1, US7661107B1|
|Inventors||Korbin Van Dyke, Paul Campbell, Don A. Van Dyke, Ali Alasti, Stephen C. Purcell|
|Original Assignee||Advanced Micro Devices, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (28), Referenced by (10), Classifications (12), Legal Events (6)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention relates generally to computer architectures and more specifically to computer architectures for processing multimedia data.
Computers traditionally developed around a single central processing unit. As more differ ways of getting information to and from the central processing unit were developed, they were attached to the central processing unit, but the central processing unit remained essentially unchanged.
Such prior art computer architectures suffer from inflexibility in the allocation of resources. For example, when performing processing that does not involve processing of audio data, audio subsystem 103 remains unused. It cannot be used for other purposes since it is permanently configured specifically for processing audio data. Thus, the overall processing capability of the computer architecture is limited by the processing capability of processor 101.
Attempts in the prior art to avoid the inefficiency of the separate application-specific communications subsystem 102, application-specific audio subsystem 103, and application-specific video subsystem 104 have used a single combined subsystem to perform the functions of subsystems 102, 103, and 104. However, such single combined subsystems still suffer from inflexibility with respect to allocation of resources between the single combined subsystem and processor 101.
Likewise, the processing capability of processor 101 is also limited. For example, a typical modern microprocessor, such as a Pentium III, often executes 200-300 cycle instructions that cannot be interrupted. Also, a typical operating system for such a microprocessor, such as Microsoft Windows, can have high and unpredictable latency with limitations, such as floppy disk accesses that cannot be interrupted.
Therefore, a method and apparatus is needed to provide increased flexibility and efficiency in the allocation of resources in a computer system.
The invention provides a method and apparatus for dynamic allocation of processing resources. By using a plurality of processors designed to have low and predictable interrupt latency that is capable of being configured to efficiently support multimedia processing without being encumbered by the Windows operating system while also being capable of being configured to efficiently support the Windows operating system, one embodiment of the invention increases the flexibility and efficiency of allocating system resources.
In accordance with one embodiment of the invention, a single integrated circuit design may be provided to support both general-purpose processing and multimedia processing. This obviates the need to design a separate device for each function to be performed by a computer system. Several instances of the single integrated circuit design may be provided in a system to increase the capability of the system. Processing resources can be efficiently allocated among tasks. Changes to the configuration of the system can be effected by changes in software, without the need for replacement of or addition to hardware.
Input/output interface 308 is coupled to bus 201. A first input/output device 309, which might, for example, be a video device, is coupled to input/output interface 308. A second input/output device 310, which might, for example, be an audio device, is coupled to input/output interface 308. A third input/output device 311, which might, for example, be a CD or DVD disk device, is coupled to input/output interface 308.
Processors 202, 203, 204, and 205 may be provided with a set of instructions in a program to cause them to perform different functions. For example, processor 202 may function as a general-purpose processor, while processor 203 may be provided with instructions to cause it to function as a video processor. As a video processor, processor 203 can communicate with a video device, for example, first input/output device 309 via bus 201 and input/output interface 308.
As another example, processor 204 may be provided with instructions to cause it to function as an audio processor. As an audio processor, processor 204 can communicate with an audio device, for example, second input/output device 310 via bus 201 and input/output interface 308.
Likewise, as another example, processor 205 may be provided with a set of instructions in a program to cause it to function as a CD or DVD processor. As such, processor 205 can communicate with a CD or DVD device, for example, third input/output device 311 via bus 201 and input/output interface 308. U.S. patent application Ser. No. 09/427,168, filed Oct. 26, 1999, entitled Transferring Execution from One Computer Instruction Stream to Another, is incorporated herein by reference.
Processors 202, 203, 204, and 205 can perform memory operations on memory device 307 via bus 201 and memory interface 306. While four processors are illustrated, the invention may be practiced with any number of processors.
The system may be initialized by having one of processors 202, 203, 204, or 205 obtain instructions from a known memory location. The instructions determine how the processors should be configured initially. The system may be subsequently reconfigured if desired.
Processing resources 401, 402, and 403 represent capability for processing tasks 404, 405, and 406. This capability may include processors 202, 203, 204, and 205. A portion of processors 202, 203, 204, and 205 may be reserved for other uses and not included within processing resources 401, 402, and 403.
Kernel 407 quantifies the capability of processing resources 401, 402, and 403, quantifies the processing requirements of tasks 404, 405, and 406 subject to any constraints, priorities, and dependencies that may exist, and dynamically allocates tasks 404, 405, and 406 to processing resources 401, 402, and 403.
Thus, kernel 407 according to one embodiment of the invention can be considered to provide a method for dynamically allocating processing resources for among a plurality of tasks, for example including multimedia tasks, or a method for dynamically allocating a plurality of tasks to a plurality of processors, which may be identical processors.
Kernel 407 may include portions expressed in different instruction sets, for example, a native instruction set and an x86 instruction set. These different instruction sets may include a first instruction in a first instruction set that is denoted by a particular bit pattern and a second instruction in a second instruction set that is denoted by the same particular bit pattern. However, the first instruction in the first instruction set may denote a different operation to be performed than the operation to be performed as denoted by the second instruction of the second instruction set.
Kernel 407 may coordinate different tasks expressed in different instruction sets. For example, task 404 may be expressed in a native instruction set, while task 405 may be expressed in an x86 instruction set. Kernel 407 allows resulting data that results from the performance of one task expressed in one instruction set to be used as initial data in the performance of another task expressed in another instruction set. U.S. patent application Ser. No. 09/239,194, filed Jan. 28, 1999, entitled Executing Programs for a First Computer Architecture on a Computer of a Second Architecture, is incorporated herein by reference.
After time T1, the first processing resource P1 is available to be allocated for other tasks. At time T4, the first processing resource P1 is allocated to begin processing of a fourth process 504. At time T2, the second processing resource P2 becomes available to be allocated for other tasks and is allocated to assist in the processing of the fourth process 504. At time T3, the third processing resource P3 becomes available to be allocated for other tasks and is allocated to assist in the processing of the fourth process 504. Thus, after time T3, first processing resource P1, second processing resource P2, and third processing resource P3 are all allocated to processing fourth process 504.
The invention may also be used to divide the processing capability of a single processor between multiple tasks. Thus, if two tasks are being performed in a system with three processors, the first processor could be assigned to perform the first task, the second processor could be assigned to perform the second task, and the processing capability of the third processor could be divided to assist the first processor in performing the first task, and the second processor in performing the second task.
Processing resources may be obtained by dividing or aggregating the processing capability of one or more processors. For example, the processing capability of a single processor may be divided in space or time to provide a processing resource. Dividing the processing capability in space refers to using some amount of the processing capability of a processor to provide one processing resource while simultaneously using other amounts of the processing capability of the processor to provide other processing resources. Dividing the processing capability in time refers to using the processing capability of a processor for some period of time to provide one processing resource and using the processing capability of the processor for a different period of time to provide another processing resource. The periods of time may be alternated or cycled between processing resources. Also, division of processing capability in space and time are not mutually exclusive, so processing capabilities may be divided in both space and time to provide processing resources.
Processing resources may be obtained by combining the processing capabilities of more than one processor. For example, the processing capabilities of two processors may be combined to provide a processing resource. The processing capabilities of multiple processors need not be combined in integer quantities. Rather, the processing capability of a processor may be divided in space or time and combined with the processing capability, or portions thereof, of another processor to obtain processing resources. The ability of processing resources to be obtained by dividing and/or aggregating the processing capabilities of more than one processor allow a portion of a plurality of processors to be used for one purpose and another portion of the plurality of processors to be used for another purpose. The portion of the plurality of processors may include processing capability of the plurality of processors combined and/or divided in space and/or time.
Moreover, the internal architecture of the x86 microprocessor is not optimized for efficient processing of multimedia data. Therefore, additional application-specific circuits are needed in conjunction with the x86 microprocessor. These application-specific circuits cannot be adapted to provide the same functionality as the x86 microprocessor. Thus, the x86 microprocessor and the application-specific circuits are immutably configured to perform their pre-determined roles, their processing capabilities cannot be adjusted or re-allocated.
When the input/output kernel 703 is executed on processor 705, the input/output kernel 703 and the processor 705, taken together, provide the same functionality as application-specific circuits, such as circuits for supporting a video, audio, or CD/DVD subsystem and may be used instead of such circuits. Thus, depending on whether input/output kernel 703 or x86 kernel 704 is selected, processor 705 may be dynamically assigned to function in place of a general-purpose microprocessor or application-specific circuits.
The input/output kernel 703 or x86 kernel 704 keeps track of available processing resources and can instruct a processing resource to start executing code at a specific address. The kernels 703 and 704 can also utilize a look-ahead capability to forecast and meet task deadlines associated with particular tasks.
Since x86 kernel 704 operates below operating system 702, x86 kernel 704 can manipulate memory used by operating system 702. For example, x86 kernel 704 can take a portion of memory that would otherwise be allocated to operating system 702 and reserve that portion of memory. Likewise, input/output kernel 703 can control or adjust memory allocation.
The invention may be practiced using apparatus comprising a plurality of processors coupled to a bus, an input/output interface coupled to the bus, and a plurality of input/output devices coupled to the input/output interface. The plurality of processors process program code configured to perform a plurality of tasks The program code comprises program code configured to cause a first portion of the plurality of processors to interact with a first input/output device of the plurality of input/output devices. The program code also comprises program code configured to cause a second portion of the plurality of processors to interact with a second input/output device of the plurality of input/output devices. The program code further comprises program code configured to cause a second portion of the plurality of processors to emulate a specific microprocessor instruction set. The apparatus may further comprise kernel program code configured to dynamically allocate the processing of the program code among the plurality of processors.
It can be readily appreciated that the input/output devices may be of a bidirectional nature. For example, processing resources may receive initial data from an input/output device, perform tasks to produce resulting data, and pass the resulting data to the same or a different input/output device.
Other software applications, for example Windows applications 904 are supported by Windows dynamic link libraries 905, which, in turn, are supported by Windows drivers 906, which are programmed in native code. Alternatively, Windows applications 904 may bypass the Windows dynamic link libraries 905 and be supported directly by Windows drivers 906. Windows drivers 906 are supported by kernel host library 903.
Kernel host library 903 is supported by kernel 910. Kernel 910 also supports graphic process 907, audio process with threads 908, and video decoder 909. Kernel host library 903, kernel 910, graphic process 907, audio process with threads 908, and video decoder 909 are programmed in native code, as denoted by being enclosed within region 911.
Thus, a method and apparatus for dynamic allocation of processing resources has been provided.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4597041 *||Nov 15, 1982||Jun 24, 1986||Data General Corp.||Method and apparatus for enhancing the operation of a data processing system|
|US5283900 *||Oct 2, 1989||Feb 1, 1994||Spectron Microsystems, Inc.||Real-time operating system and virtual digital signal processor for the control of a digital signal processor|
|US5526521 *||Jan 25, 1995||Jun 11, 1996||International Business Machines Corporation||Method and system for process scheduling from within a current context and switching contexts only when the next scheduled context is different|
|US5706514 *||Mar 4, 1996||Jan 6, 1998||Compaq Computer Corporation||Distributed execution of mode mismatched commands in multiprocessor computer systems|
|US5724587 *||Jun 8, 1995||Mar 3, 1998||International Business Machines Corporation||System for controlling task execution in a host processor based upon the maximum DMA resources available to a digital signal processor|
|US5841763 *||Jun 13, 1995||Nov 24, 1998||Multilink, Inc.||Audio-video conferencing system|
|US5913058 *||Sep 30, 1997||Jun 15, 1999||Compaq Computer Corp.||System and method for using a real mode bios interface to read physical disk sectors after the operating system has loaded and before the operating system device drivers have loaded|
|US5991808 *||Jun 2, 1997||Nov 23, 1999||Digital Equipment Corporation||Task processing optimization in a multiprocessor system|
|US5999990 *||May 18, 1998||Dec 7, 1999||Motorola, Inc.||Communicator having reconfigurable resources|
|US6078944 *||Apr 1, 1997||Jun 20, 2000||Hitachi, Ltd.||Process management method and system|
|US6098091 *||Dec 30, 1996||Aug 1, 2000||Intel Corporation||Method and system including a central computer that assigns tasks to idle workstations using availability schedules and computational capabilities|
|US6125394 *||Jun 6, 1997||Sep 26, 2000||At&T Corporation||Computer system having a plurality of resources and utilizing a selection mechanism to select the resources based upon historical loading|
|US6173306 *||Jul 21, 1995||Jan 9, 2001||Emc Corporation||Dynamic load balancing|
|US6289369 *||Aug 25, 1998||Sep 11, 2001||International Business Machines Corporation||Affinity, locality, and load balancing in scheduling user program-level threads for execution by a computer system|
|US6292822 *||Jun 30, 1998||Sep 18, 2001||Microsoft Corporation||Dynamic load balancing among processors in a parallel computer|
|US6301603 *||Feb 17, 1998||Oct 9, 2001||Euphonics Incorporated||Scalable audio processing on a heterogeneous processor array|
|US6338130 *||Mar 11, 1999||Jan 8, 2002||International Business Machines Corporation||Adaptive method and apparatus for allocation of DSP resources in a communication system|
|US6381321 *||May 4, 1998||Apr 30, 2002||T-Netix, Inc.||Telecommunication resource allocation system and method|
|US6446129 *||Jun 25, 2001||Sep 3, 2002||Emc Corporation||Method and apparatus for synchronizing function values in a multiple protocol system|
|US6493776 *||Aug 12, 1999||Dec 10, 2002||Mips Technologies, Inc.||Scalable on-chip system bus|
|US6496922 *||Oct 31, 1994||Dec 17, 2002||Sun Microsystems, Inc.||Method and apparatus for multiplatform stateless instruction set architecture (ISA) using ISA tags on-the-fly instruction translation|
|US6549945 *||Mar 11, 1999||Apr 15, 2003||International Business Machines Corporation||Multi-purpose WAN driver for DSP resource adapter|
|US6658448 *||Oct 21, 1999||Dec 2, 2003||Unisys Corporation||System and method for assigning processes to specific CPU's to increase scalability and performance of operating systems|
|US6665701 *||Aug 3, 1999||Dec 16, 2003||Worldcom, Inc.||Method and system for contention controlled data exchange in a distributed network-based resource allocation|
|US6938256 *||Jan 18, 2001||Aug 30, 2005||Galactic Computing Corporation||System for balance distribution of requests across multiple servers using dynamic metrics|
|US7196708 *||Mar 31, 2004||Mar 27, 2007||Sony Corporation||Parallel vector processing|
|US7376812 *||May 13, 2002||May 20, 2008||Tensilica, Inc.||Vector co-processor for configurable and extensible processor architecture|
|US20020016811 *||Oct 4, 2001||Feb 7, 2002||International Business Machines Corporation||Computer system and method for sharing a job with other computers on a computer network using IP multicast|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8296490 *||Jun 29, 2007||Oct 23, 2012||Intel Corporation||Method and apparatus for improving the efficiency of interrupt delivery at runtime in a network system|
|US8381223||Jul 25, 2011||Feb 19, 2013||Korbin Van Dyke||Method and apparatus for dynamic allocation of processing resources|
|US8838864||Oct 23, 2012||Sep 16, 2014||Intel Corporation||Method and apparatus for improving the efficiency of interrupt delivery at runtime in a network system|
|US9311157||Sep 27, 2011||Apr 12, 2016||Samsung Electronics Co., Ltd||Method and apparatus for dynamic resource allocation of processing units on a resource allocation plane having a time axis and a processing unit axis|
|US20090007150 *||Jun 29, 2007||Jan 1, 2009||Yadong Li||Method and Apparatus for Improving the Efficiency of Interrupt Delivery at Runtime in a Network System|
|US20140043344 *||Mar 16, 2012||Feb 13, 2014||Sunil A. Kulkarni||Techniques for a secure graphics architecture|
|CN104050007A *||Mar 14, 2014||Sep 17, 2014||英特尔公司||QOS based binary translation and application streaming|
|EP2622474A2 *||Sep 26, 2011||Aug 7, 2013||Samsung Electronics Co., Ltd||Method and apparatus for dynamic resource allocation of processing units|
|EP2622474A4 *||Sep 26, 2011||Apr 16, 2014||Samsung Electronics Co Ltd||Method and apparatus for dynamic resource allocation of processing units|
|EP2825952A4 *||Mar 16, 2012||Oct 14, 2015||Intel Corp||Techniques for a secure graphics architecture|
|U.S. Classification||718/104, 345/506, 712/36, 345/502, 712/35, 718/102|
|International Classification||G06F9/46, G06F15/76, G06F1/20, G06F15/16|
|Jan 18, 2000||AS||Assignment|
Owner name: ATI INTERNATIONAL, SRL,BARBADOS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:VAN DYKE, KORBIN;CAMPBELL, PAUL;VAN DYKE, DON;AND OTHERS;SIGNING DATES FROM 19991217 TO 20000104;REEL/FRAME:010553/0939
|Nov 30, 2009||AS||Assignment|
Owner name: ATI TECHNOLOGIES ULC,CANADA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ATI INTERNATIONAL SRL;REEL/FRAME:023574/0593
Effective date: 20091118
Owner name: ATI TECHNOLOGIES ULC, CANADA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ATI INTERNATIONAL SRL;REEL/FRAME:023574/0593
Effective date: 20091118
|Dec 22, 2009||AS||Assignment|
Owner name: ADVANCED MICRO DEVICES, INC.,CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ATI INTERNATIONAL, SRL;REEL/FRAME:023690/0069
Effective date: 20091222
|Aug 31, 2010||CC||Certificate of correction|
|Nov 14, 2012||AS||Assignment|
Owner name: ATI TECHNOLOGIES ULC, CANADA
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ERRONEOUS RECORDATION AND REQUEST REMOVAL OF PAT. APPL. NO. 09/484,549 (PAT NO. 7,661,107) PREVIOUSLY RECORDED ON REEL 023574 FRAME 0593. ASSIGNOR(S) HEREBY CONFIRMS THE ASSIGNMENT OF ASSIGNOR S INTEREST EXCEPT FOR PAT. APPL. NO. 09/484,549 (PAT NO. 7,661,107);ASSIGNOR:ATI INTERNATIONAL SRL;REEL/FRAME:029300/0313
Effective date: 20091118
|Mar 18, 2013||FPAY||Fee payment|
Year of fee payment: 4